Archives nobody can search
A useful moment exists, but nobody knows which video it is in, or at which minute.
Content capital · Technical sheet
Find the right moment in your videos, with its source.
From an authorised corpus, Aive Expert organises queryable content units and points back to the available sources and moments that support its answer. Coverage depends on the content ingested, its quality, the analyses enabled and the rights.
Last verified · 26 September 2026
What do Aive video indexing and search do?
Aive organises an authorised video corpus into queryable units: transcript, speakers, chapters, timecodes and metadata, depending on available analyses. A question points to the relevant moments, with source and timecode. When Aive Expert is enabled, search works on intent rather than on file names. Results must still be checked against the source: neither completeness nor an answer is promised.
- Content unit
- A queryable segment of a video (moment, chapter or answer) linked to its source and timecode.
- Authorised corpus
- All content your workspace is entitled to ingest and query.
- Timecode
- The timestamped position of a moment in the source video.
The operational problem
The proof exists. Nobody can find it.
Search by file name
Titles and folders describe neither the topics covered, nor the speakers, nor the proof.
Quotes without a source
A moment reused without timecode or version can be neither verified nor reused again.
Mechanism
From video file to the moment you need.
01Inputs
- Authorised videos in the workspace
- Existing transcripts and subtitles
- Metadata and associated rights
- Questions or queries from your teams
02Processing
- Ingestion and quality control
- Transcript, speakers and chapters depending on available analyses
- Content units linked to timecodes
- Corpus index when the module is enabled
03Outputs
- Relevant moments with source and timecode
- Sourced answers from Aive Expert
- Missing data flagged
- Downstream reuse subject to configuration
Technical sheet
What Aive Expert covers.
Vectorisation and semantic search are described here only within the scope of an enabled Aive Expert (Media Factory). Any other use is to be confirmed during scoping or in the contract.
| Item | Detail | Status |
|---|---|---|
| Module | Media Factory, with Aive Expert enabled in your workspace. Indexing alone, outside Expert, is to be validated. | Available / to be validated by corpus and subscription |
| Corpus | Authorised content ingested into the workspace; the covered corpus is displayed. | Depending on your workspace |
| Modalities | Transcript, chapters, speakers, metadata and timecodes, depending on the analyses enabled. Any other modality only if demonstrated and contracted. | To be validated per module |
| Mechanism | Vectorisation of content units and semantic search across the corpus, to find a moment by intent. | To be validated per module |
| Outputs | Moments, sources, timecodes and sourced answers pointing back to the corpus. | Subject to configuration |
| Security | Isolated workspace, roles, quotas and revocation; customer data is never used to train models. | Subject to configuration |
| Limits | Coverage depends on the content ingested, its quality, the analyses enabled and the rights; answers must be checked and are neither complete nor guaranteed. | Always apply |
| Entitlement | Defined by subscription or contract: corpus, quotas and access. | Not public · per contract |
Sheet checked on 26 September 2026. No capability described here is attributed to MGT without a dedicated product sheet.
Demonstration
One question, one sourced moment.
Demo data. The component illustrates the question, moment, source journey; it does not reflect your corpus.
Aive Expert · demo corpus Demo data
How do I prepare a video for GEO?
Start with the transcript and chapters, then link every important statement to its timestamped moment. A structured page with VideoObject and a direct answer makes the video easier to understand and verify.
[1]Video GEO masterclass · moment 04:12
[2]Structured data guide · moment 01:37
Answer depends on the authorised corpus and available sources: verify in the source. What is a sourced statement?
A statement linked to the proof that supports it: a video moment, a document or primary data, with its timecode and version. The statement stays reviewable by the team before publishing.
[1]Content Graph explained · moment 02:48
Answer depends on the authorised corpus and available sources: verify in the source. Are AI citations guaranteed?
No. No tool can promise that an engine will cite a piece of content. The demo corpus explains that measurement relies on observed answers, within a defined panel and period.
[1]Measurement methodology · moment 00:55
Answer depends on the authorised corpus and available sources: verify in the source. Control
Measured, controlled, scoped.
What is measured
- Coverage of the ingested corpus
- Missing or low-quality data
- Moments returned and their sources
- Questions with no answer in the corpus
Controls
- Authorised, displayed corpus
- Asset version and provenance
- Roles and access per workspace
- Revocation of content from the corpus
Prerequisites and availability
- Ingestion rights over the content
- Aive Expert (Media Factory) enabled for semantic search
- Status to be validated per module
- Scope confirmed during scoping or in the contract
Frequently asked questions
What is the difference between indexing and semantic search?
Indexing organises the corpus: transcript, speakers, chapters, timecodes and metadata. Semantic search, when Aive Expert is enabled, finds a moment by intent rather than by exact wording. Both depend on the content ingested and the analyses available.
Are these capabilities part of MGT?
This page describes the scope of Aive Expert, part of Media Factory. No indexing or search capability is attributed to MGT without a dedicated product sheet; the scope is confirmed module by module.
Are results complete?
No. Coverage depends on the content ingested, its quality, the analyses enabled and the rights. Every answer points to a source that must be checked.
Are my videos used to train models?
No. We never use our business customers’ data to train our models. The corpus stays attached to your workspace and content can be revoked from it.
Test search on a defined corpus.
In a demo, try search on your content; scoping then sets a corpus, its rights and the questions to test.